Benchmarking: What's Enough?

I’ve had a lot of questions about benchmarking the size of your affiliate network. What’s enough? Affiliate marketers want to know: Am I doing enough? Do I have enough affiliates? Am I growing fast enough? Am I driving enough revenue from my affiliate program? Do I have enough super affiliates?

The fact is that even with more than 3,400 merchants (source: CashPile) now running affiliate marketing programs, data is scant. The last Forrester and Jupiter studies on affiliate marketing were released in the fall of 1999. Information sharing by participants and networks alike is almost nonexistent. And getting your hands around the size of the web universe can be difficult.

Forrester’s nearly one-year-old report on affiliate marketing pegged the size of the average affiliate network at just more than 10,000 affiliates. But is that big enough? One of my favorite benchmarks of web site growth is Netcraft’s Web Server Survey. While the survey is really intended to chronicle which web sites are using which web server software, it’s also useful for plotting the growth of the web. Specifically, Netcraft collects and collates as many hostnames connected to the Internet as it can find and systematically polls each one.

Sizing the Web

Basically, Netcraft counts the total number of web sites. In July 2000, Netcraft found 18.2 million total web sites, up from 17.1 million web sites in June. As another data point, the 18.2 million web sites found compares favorably with the 20.6 million domains that DomainSurfer currently reported as registered on August 23, 2000. (Note: If you haven’t tried it, DomainSurfer is a handy utility for meta searching all competing domain registrars simultaneously, including very powerful wildcard searches.)

Thankfully, Netcraft’s primary count excludes personal home pages at sites like Angelfire, Tripod, HomePage, Homestead, GeoCities, TheGlobe, and Fortune City. For the record, a review of company press releases and SEC filings shows that personal home pages represented approximately 34 million “sites” at the close of Q1 2000 again, all of which are purposely excluded from Netcraft’s counting.

Does Size Matter?

But this begs more questions. What about people who buy domains and never build web sites? For example, at the time of the June survey, Netcraft reported that Register.com hosted roughly 1.4 million domain names, the great majority of which were template sites. And what about places like Freeserve where roughly 170,000 sites are hosted on only four load-balanced IP addresses most of which are active sites produced by real people?

Active Sites Are What Count

How many active sites are there on the web? To make its numbers more useful and answer this question, Netcraft added a new methodology starting with the June 2000 survey that counts only sites with unique content. Simply stated, Netcraft takes the front page from each site and compares it to the front page of other sites with the same IP address. For services that host large numbers of sites, Netcraft has developed a method by which the number of pages fetched is logarithmically proportional to the number of sites on any given IP address. According to Netcraft, this ensures that even in the most pathological case, the survey is not banned from sites and that the results remain respectable and accurate. I encourage you to read Netcraft’s full methodology, but here’s my take on the stuff that’s useful to marketers.

Detecting Placeholder Domain Registrations

First, Netcraft’s new “active” methodology eliminates sites that are merely placeholder domain registrations. For example, the biggest domain registries are large enough to be significant even in the context of the 17 million sites found by the June 2000 Web Server Survey. Register.com and Network Solutions alone accounted for about 12 percent of the sites found in the June 2000 Web Server Survey. Using Netcraft’s sampling technique, Register.com went from a count of 1.4 million sites to a count of about 100,000 active sites.

Removing Corporate Aliases

Next, many companies register in more than one domain but point them to the same primary web site. For example, Netcraft holds the netcraft.com, netcraft.net, and netcraft.co.uk domains and currently uses three hostnames that will revert to the Netcraft site. This means that there are nine names in the DNS that will revert to the same content: www.netcraft.com, netcraft.com, ssl.netcraft.com, www.netcraft.net, netcraft.net, ssl.netcraft.net, www.netcraft.co.uk, netcraft.co.uk, and ssl.netcraft.co.uk. For the purposes of counting the number of active web sites, Netcraft’s new “active” methodology would count only the Netcraft site once no matter how many domains and hostnames point at the site.

Measuring Virtual Hosts

Finally, Netcraft now accurately measures services like Freeserve, where about 136,000 of approximately 170,000 sites are shown to be active. In the words of Netcraft, “This is a nice test case for the methodology, as Freeserve is a large number of active sites on just four IP addresses using HTTP/1.1 virtual hosting.” Exactly.

The Results

Of the more than 17 million sites found by the June 2000 Netcraft survey, nearly 10 million were removed from the survey by applying the active sites methodology. Essentially, the domain name registries disappear. Cheap and free bulk hosters represent a larger share of the active sites. Larger co-location companies, such as Exodus, which may provide connectivity for “downstream” bulk hosters, also gain share of active sites. This is exactly as one would predict.

Overall, Netcraft found 7.5 million active sites on 3.4 million IP addresses during the June 2000 survey. Of the IP addresses with active sites, approximately 3.2 million have just a single active site on them. And for what it’s worth, the largest number of active sites on a single IP address was 286,620 at WebJump. WebJump was acquired by TheGlobe.com in December 1999.

Which brings us full circle. How many affiliates should you have? With 7.5 million active web sites in June 2000, I’ll leave you to draw your own conclusions. But my thinking tells me that 10,000 affiliates is nothing but a rounding error. Let’s all think bigger. Tell me what you think: Do you want more numbers like these?

Shameless Self-Promotion

I invite you to visit the beta version of my start-up EchoFactor.com. EchoFactor is a content traffic exchange, driving traffic to top Media Metrix web sites by offering free headline feeds to affiliate webmasters. Our beta version already covers more than 14,000 sports, entertainment, college, and regional topics. What affiliate networks did for commerce sites like Amazon.com and Art.com, EchoFactor does for content sites like CNET, ESPN, and VerticalNet. If you run a content web site and aren’t getting enough traffic, I invite you to email me directly. Since we’re in beta, I also welcome any and all comments and criticisms.